Global and Local Facial Feature Extraction using Gabor Filters

Recently face recognition is utilized in real time applications like security systems, attendance systems, video-surveillance systems, visitor management systems, etc., Face recognition systems are mainly composed of two modules: face detection module and face verification module. The purpose of former is to determine whether there are any faces in an image, while the later involves confirming or denying the identity of that person. Face recognition algorithms commonly assume that face images are well aligned and have a clear pose, yet it is not possible in many practical applications. The features of the face are extracted using two filters, i.e., local filters focused on accuracy and global filters focused on processing time. Real time applications require both these features. The paper proposes a face recognition system which involves Gabor filters, that are used for extraction of the global and local facial features. These local features are combined when the global features represented are not clear. These features are fused to form a feature vector to be used as a face descriptor for recognition. Gabor filter is pose robust and this filter works with sub blocks and hence it is highly illumination tolerant. The fusion of local and global features gives good representation for recognition. 2Dimensional Hidden Markov Model (2D-HMM) is used for recognition of the faces from the database.